摘要： 提出了一种能够快速进行复杂形状多边形匹配的算法，该算法基于正切空间表示，先对复杂多边形进行离散曲线演化，再将得到的简化多边形分为一系列最大凸/凹弧线，并选择每段最大凸弧线的起点作为匹配的起始点进行匹配。实验结果证明该算法不但能够对复杂多边形快速而精确的匹配，而且具有不受噪声影响的优点。

关键词: 形状匹配, 相似性度量, 离散曲线演化

Abstract: Shape matching is an important issue in pattern recognition, and polygon-based shape representation is a common method used in shape description. This paper presents a method for quick matching of complex polygons based on tangent space representation. The complex polygon is first evolved into a simplified one using discrete curve evolution, then is divided into a serial of maximum convex/concave arcs. The beginning point of each maximum convex arc is used as the reference point of matching. The experimental results justify that the shape matching method is efficient and is stable with respect to noise distortions.

中图分类号: 

• TP391.41